Monday, October 24, 2011

Crabaugh's critique of classes

Dragon #109 (May 1986) contains one of my top 10 all-time favorite articles from that venerable magazine.  "Customized Classes" by Paul Montgomery Crabaugh.  I've sung the praises of Crabaugh before.  Dude wrote only a handful of published articles but each were brief and focused, with a strong grasp of the mechanics he was deploying.

Which is one of the reasons why "Customized Classes" is so interesting.  In just five pages Crabaugh gave us a system for making our own BX classes, demonstrated how it worked by redoing all seven canonical classes, and supplied us with five new classes to show how his system is meant to be used.  Kick ass.

But one thing has bothered me about this article for 25 years now: Crabaugh's numbers for the seven BX classes don't add up to the totals in the rulebook.  Dig these numbers for achieving 2nd level.

Fighter: rulebook 2,000xp; Crabaugh 1,760xp
Magic-User: rulebook 2,500xp; Crabaugh 1,840xp
Cleric: rulebook 1,500xp; Crabaugh 2,160xp
Thief: rulebook 1,200xp; Crabaugh 1,460xp
Dwarf: rulebook 2,200xp; Crabaugh 1,840xp
Halfling: rulebook 2,200xp; Crabaugh 1,560xp
Elf: rulebook 4,000xp; Crabaugh 2,780xp

For a long time I thought that the dude just did his level best to hit the numbers in the rulebook but only came close.  Two things have turned me around on this reading in the past four or five years.  First of all, every other Crabaugh article I've encountered (not that there are many) indicates that the dude was on the ball with the rules, suggesting to me that he didn't just eyeball his numbers and hope for the best.  Secondly, Crabaugh's version of the cleric gets a spell at first level.  You don't write a BX article and give 1st level clerics a spell unless you are critiquing the rules as written.  (I guess you could do that if you're a cheap hack and not really paying attention to the edition you're writing for, but those kind of guys generally wrote AD&D articles during this period.)

So what do Crabaugh's numbers tell us about his opinions of the various classes, at least with regards to their first level version?  Here's what I think he's saying:
  • The vast 500 point difference between fighters and MUs is not justified.
  • Thieves advance too fast, but they are still wimpy enough that they should advance faster than the other human classes.
  • Halflings are not nearly as useful as dwarves or fighters.  I love halflings but I think I agree.
  • Elves are clearly the most potent class in the game, but they aren't worth nearly as many XP as the BX rules suggest.
  • IF you give the cleric a spell at first level it is the most potent human class and not that much less powerful than an elf.
 What does everyone think about Crabaugh's numbers?  I think a lot of players will agree that the MU has a big hill to climb.  Should the elf XP requirements be reconsidered as well?


  1. He is on the money, especially with clerics and elves. Both are in effect a combined fighting and magic-using class (especially with cleric spells at 1st level, which is a good move) so both should command a premium, but 4000 for the elf always seemed kind of like overkill.

  2. Anonymous4:54 AM

    I agree more with his numbers than the official ones. I've always felt some of those totals were off in the game, and do not believe the original rules/authors are infallible. Kind of the opposite from the idea that Crabaugh eyeballed it, I think the authors of the rules did, and then maybe playtested to see if it felt right. I suspect he reasoned it out and used math better, and his totals are closer to what I would suggest, based on years of play.

    As for the cleric spell at 1st level, I can see why not having that might be a good thing or make sense, but it seems everyone who plays a cleric wants it and is surprised when you tell them they don't get it till second level. Without it they are almost like blunt weapon only fighters with a turn undead ability added. It's also tough to have a character who will be able to heal at 2nd level when everyone at 1st level is dying cause they don't have that ability.

    So I guess I am saying... I like giving them the spell at 1st, and increasing the XP a lot. Clerics *should* need more xp to advance, even without the spell, in my opinion.

  3. Clerics with a sell (any 1st level cleric spell) at 1st level are certainly the toughest of the classes.

    Elves are fine kept in check with the high exp requirements to me.

  4. I don't think I was aware there were different XP requirements for different classes... interesting to consider.

  5. Great article that. He seems to be in the right ballpark XP-wise: clerics and thieves were always somewhat undercosted IMO.

    Reading the 2E DMG the other day, it looks like the Crabaugh Method for creating new classes was later canonised as an "official AD&D©®™ pat.pending optional rule" (pp22-23, entirely uncredited, and right before 6 (yes, six!) eminently forgettable pages on Alignment).

  6. Chris, the version in the 2nd edition DMG has flaws you won't find in Crabaugh's version. I'm pretty sure that under those rules you can make a functional (albeit very low-powered) class that needs 0xp for level 2.

  7. A Cleric with a spell is hands down the toughest class at 1st level. I agree with everyone else who feels Crabaugh's numbers seem more reasonable.

    However, I have to admit preferring nice even numbers like 2,000 over 1,760. :) Maybe round everything up to the next highest 100?

    Thief: 1,500
    Halfling: 1,600
    Fighter: 1,800
    Dwarf: 1,900
    Magic-User: 1,900
    Cleric: 2,200
    Elf: 2,800

  8. I've had a hard time with Crabaugh's approach, essentially because you can't use it to build the existing classes. Whatever you create won't balance against the canon.

    Self-promotion time: I use this method of class building. It's based on reverse-engineering the existing classes, so the XP balance is built-in.

  9. No need to overstate your case, Erin. Your system works better assuming you agree that the canonical numbers represent fair XP targets. Questioning that ssumption is sort of the whole point of my post today.

    Though you seem to have one of the same flaws as the 2e DMG version. If I build a class with d4 HD, save as Fighter, fight as MU, no armor and 'restricted' weapons, then it looks like I need 0 XP to reach 2nd level. By adding a level limit and/or dropping all weapon use I can actually add some useful abilities. Unless I reading it wrong, under your system I could easily whip up a non-violent healer type that flies up the level chart.

  10. Weird. While I'm not going to comment on clerics since the author of the article has already modified them, I have some issues with the elf and magic-user progression.

    In my opinion, a Magic-User doesn't loose out on much on being an elf. Trade in 1 spell per day for a better hit die, two steps up on the to-hit progression, immunity to ghoul paralysis, the ability to use armor and any weapon? Yes please.

    The problem with the demi-human classes (except Halflings, I think those are perfect as is) is that they don't give up anything at first level in exchange for special racial abilities compared to the base class. Yes clerics have armor and spell casting ability, but no spells at first level - a good trade off in my opinion. There is nothing else to even this out for the elf and the magic-user except the enormous XP requirement.

    My verdict: keep as is.

  11. I definitely prefer Crabaugh's numbers. Even without a spell at first level a cleric is pretty powerful, though less so than it is with the spell. His are also almost universally lower, which I think most players would appreciate.

    I've thought about using them before, but I'm typically a bit wary of adding house rules. As far as making new classes go, I find that his method for adjusting AD&D classes in the same article works just as well with BX/CMI. The numbers come out much closer to the canon ones that way.

  12. When odds makers set the odds, they often do not set them according to what they believe the odds actually are but what they believe other people think they are.

    It might be true, say, that magic-users are not 20% better than fighters (2,500 vs. 2,000) or 67% better than clerics (2,500 vs. 1,500) at first level, but if you didn't give them such a high experience threshold, then perhaps an additional unhealthy number of players would want to play them.

    In other words, experience level differences may be used to counterbalance subjective attractiveness, not merely objective value.

  13. You know, that’s interesting. I don’t know why I always discounted these kinds of systems for not reproducing the canonical progressions. I would agree the these numbers are probably better than canon.

    Although, I still suspect there’s something of an X factor that gets missed in an analytical approach. Still, it seems worth giving Crabaugh’s system a try and see how it goes. With some tweaks based on trying it, I’m betting it could made to fit just about anyone’s group.

  14. I prefer to think the xp situation isn't based on effectiveness but personality type:

    -Anyone playing a magic user from level 1 clearly has his or her eyes set on glittering vistas of fireballs and walls of ice. You can make him or her do almost anything because this player is hungry for power s/he will one day get if s/he keeps playing. So why not force him to do twice as much work to level up? It'll make whoever's playing that PC call their friends who are passing them by in levels and be like "c'mon, let's play, I need xp".

    -Anyone playing a thief is fucking hardcore as fuck. This person will be playing D&D forever, probably, because why else would you pick a class where you basically can't do anything and you're just "world flavor"? Because you're -really into- it. So let the thief level up. Reward loyalty to the product. So what if he does? it's not like anything happens once he's levelled up anyway.

    -Cleric rewards should be disproportionately vast otherwise who would ever play them?

    -The fighter: this is the casual gamer--you want to make it easy enough for the fighter to level up that they lear what that us like and that it's fun but hard enough that s/he wants to keep playing.

    -The elf: like the magic user, your elf player lusts for power, unlike you magic user, your elf player will definitely use that power to hog the spotlight and ruin everyone else's game. So make it hard for this guy. Perhaps, as a "hey let's play so i can level" engine, the elf will turn out to be more useful than s/he is scary. Use their unseemly lust to power the game, at least.
    Also, rather than think in levels like 8th level fighter v. 9the level MU, in AD&D I tend to think solely in xp numbers, like "What can an fighter with 10,000 xp do vs an MU with 10,000 xp?

  15. Never had a problem with elves requiring heaps of xp.
    On the other hand clerics should pay a lot more IMHO: Spells, Turn Undead, good armor and decent combat abilities are simply a steal those xp totals.

    The rest is just OK.

  16. No need to overstate your case, Erin. Your system works better assuming you agree that the canonical numbers represent fair XP targets.

    Well, you did ask ;)

    My premise was that any B/X compatible class-creation system (which, ostensibly, Crabaugh was trying to create) necessarily derives from the canonical XP targets. Otherwise, you're just assigning XP values that "feel" right to you, but which might not translate well from group to group (as suggested by the comments people have already made).

    Though you seem to have one of the same flaws as the 2e DMG version.

    True - your example does come out to 0XP. Though, you could do the same kind of thing with Crabaugh--if the goal is to min/max your PC's XP requirements, I don't see how any system could avoid exploitation (unless it prescribes some artificial minimum XP value).

    But on a practical level, I'd question whether or not the class you describe, with such severely constrained abilities, would be fun to play. If so, what are the benefits of level advancement to a character with no real abilities?

    Fundamentally, the real question is how much XP do you think such a class should be worth? If you're trying to balance against official classes, coming up with an arbitrary number that makes sense for level progression, but not actual ability, would seem too arbitrary to me.

    All that said, the bullet points you provide sound like the same conclusions one could make about the official numbers, even though the XP values are different. Perhaps more evidence that the "feel" of the class is more important than the math?

  17. I actually knew Paul when I was a kid, and played in his games and in many games with him because I hung out at the local game shop he frequented. Seeing as he's getting a sort of celebrithy treatment here and at Grognarida today, I'm thinking I might post a little about my experiences with the guy if I have time today. He was one of the better people among the human flotsom that hung out at Aero, so it should be a fairly kind rememberance.

  18. But on a practical level, I'd question whether or not the class you describe, with such severely constrained abilities, would be fun to play. If so, what are the benefits of level advancement to a character with no real abilities?

    Hit Dice: d6
    Saves: Fighter
    Combat: Magic User
    Armour: Restricted
    Weapons: Restricted
    Level Limit: Name Level
    Total XP for all levels: 0

    If you add d8 hit dice to give it a positive value they'll still be at level 5 with only 800xp. That's 5d8 hit points and since they're levelling up so fast it offsets the Magic User combat progression giving them the best attack values compared to the other classes.

  19. I'd be interested to know why the numbers he chose seem to be multiples of 40. Can't be an accident, right? Not having read anything by this Crabaugh dude, does he explain the significance of this? Does it crop up elsewhere in his writings?

  20. @Stuart: I'd play that! A B/X thief with better hp, and only slightly worse thieving abilities!

  21. I like Crabaugh's numbers. Heck, I modified his system for a Rules Cyclopedia game I never ran... Still have that system somewhere, I remember adding many things to it over time.

    Still wish the man was with us -- his potential influence on Traveller would have been immense.

  22. @Stuart,
    Well, yeah, if you want to make a 0 XP classe, the system lets you. But I don't think such results are really satisfying, or even viable, in practice. YMMV.

    I mean, your archer example is by-the-book, but why bother creating one when you could just play a fighter with Weapon Mastery with the bow? He'd have better to-hit and more options with his favoured weapon. Or, better yet, create an archer with some special archery abilities to differentiate him from the fighter who has Weapon Mastery with the bow?

    If you focus only on HD, save, to-hit, and armament choices, I think you're just as well off with the original classes. My thinking when I wrote the system was that you'd use special abilities, skills, and racial talents to differentiate; hence the need for a balanced approach.

  23. @Jeff - Thanks for bringing this article to my attention! I did not remember it at all. It's a very interesting read.

    I especially found the "Editor's Conclusion" to be interesting. In it he posits how the system could be used with AD&D to slightly modify existing classes. His examples are a fighter learning to pick pockets for a 10% xp penalty and a cleric using a sword for an extra 15%. I kinda like that idea.

    I could also see this concept being used in OD&D to eliminate the thief class.

  24. I love this article. I've been working on reworking this system for 2e.

    The thing that bugs me about the 2e class builder is that you are told that the classes you make with the system won't be comparable to the standard classes. Under Table 21 you find:

    You can't reconstruct the existing character classes using this method. The standard classes give players advantages over custom-designed classes. Standard class characters advance in levels more quickly and, generally, have better abilities than custom-designed characters.

    Anyway, I remember adding a lot of abilities from other classes in Dragon like the Jester and all the India-themed kits. I also added some things from the Player's Options books (need to use them for something). It was funny that when I couldn't build the Fighter up to the canon, I added the Catch/Deflect Object ability to them. The rationale was to provide a method of fighter to deflect fireball and javelins.

    To get the MU to reach canon, I had to create an entire category of making magic items. Thus, I had a sorcerer class that could cast spells, but couldn't make any magic items, including spell research.

    One thing I've noticed about any system is that if I create a Fighter type (hit dice, armors and saving throws) with limited weapons, the XP requirement per level winds up around 1200XP. I used it to make a "soldier" class that was good at following orders and had bonuses to mass combat. They could be fun with the right campaign, but not as much fun for adventuring.

  25. Anonymous6:31 PM

    Check out this thing:

    Fighter Level - XP Needed
    1 - Not a dang thing!
    2 - 2,000
    3 - 4,000
    4 - 8,000
    5 - 16,000
    6 - 32,000
    7 - 64,000
    8 - 125,000
    9 - 250,000

    The XP you need doubles every level. This is pretty standard. What EXP you needed to get all the way to name level, that's what you'll need per level thereafter.

    However, we have Druid who sputters along not doubling, then heads for Crazytown at 110 MPH at level 12 (you go from needing 100k to level up, to needing 350k, then 750k, then you hit your wall). Assassin does this too.

    Then we have the M-U, who not only starts out needing 2,500 to level up to 2, but it seems to MORE than double at level 4! Then it requires less than double, then only +50% more for the next few levels. If it had stayed with double every level, the M-U would need 1,280k per level after name level! Illusionist does the same thing 7th to 9th. Thief does it 7th to 10th. These don't correlate to big class ability changes.

    Monk does a mix of these two, offering a weird easing of XP required 8th to 10th then busts your nuts 13th to 17th. Again, nothing outstanding happens to Monk class abilities on the down side 8th to 10th or the up side 13th to 17th.

    I guess what I'm getting at is, the advancement tables in the PHB have weird quirks and making a universal XP chart or a chart based on a single XP requirement plus percentage modifier might end up weird. They seem to have been constructed to put you at a certain post-name-level XP cost, or else in classes with a level cap you get shafted at the end so you never reach it.

  26. @Erin

    I mean, your archer example is by-the-book, but why bother creating one when you could just play a fighter with Weapon Mastery with the bow?

    Because you'd like to have five hit dice and make saves on the 4-6 column when the fighter is still less than halfway to 2nd level?

    The fundamental problem with your system is that it literally counts many abilities that advance with level as worthless, rather than merely less valuable. A d4 of hp each level is obviously worth fully half of a d8 in mechanical terms, but you don't charge anything for it instead of half as much.

    What should be 0 xp are things that are truly worthless — 0 hp/level, zero save improvement/level, zero attack progression improvement. Zero-basing something on the grounds that it's the worst any class makes the bottom improve-with-level abilities overpowered for their XP cost. It makes it easy to make new classes that are not consistent with the ones existing in the game.

    And, incidentally, it would actually explain why you're getting a 900 point discrepancy on the MU XP table; it's because you're giving the MU a bunch of level-advancing abilities for free when they should cost something.

  27. The fundamental problem with your system is that it literally counts many abilities that advance with level as worthless, rather than merely less valuable.

    Fair enough, and I get your point. I just never used the approach for gaming the system, but rather mixing and matching special abilities upon a "chassis" of hit die, saves, to-hit, and weapon/armour choices, to come up with something balanced and non-arbitrary.

    Again, my original goal was to break down the original classes into individual abilities and XP values. In truth, this is the first I've seen anyone deliberately try to break the system, so my bad. I suppose a revision with non-zero values should be attempted.

  28. @Chris:

    Reading the 2E DMG the other day, it looks like the Crabaugh Method for creating new classes was later canonised as an "official AD&D©®™ pat.pending optional rule"

    No. The Crabaugh article -- as bad as it is -- at least does not have negative experience point factors in it. Those rules in the 2e DMG are infamous because you can make an all powerful God class that has conjuration spells (wish) and requires 0 EXP per level. Those were some of the worst rules ever created in all of official D&D.

    The Crabaugh article does not permit this, as everything is additive. With that said, we know that this is a poor way to design a class building system. You simply cannot capture the power of synergistic abilities with an additive mechanic.

    But then I am not aware that any of the early designers really grasped a deep understanding of the mathematics behind the rules. The earlier editions worked well from a narrative perspective, but you cannot look at the game mechanics too deeply without it falling apart.

  29. infornific9:25 PM

    There's a simple custom class template for B/X D&D at the Vault of Pandius (also seen in OD&Dities. It's deliberately set up to prevent uber - classes. One feature that I like is that elves always have full magic-user ability. I also took my own stab at a custom template here, designed to match Labyrinth Lord numbers. Personally I think magic users are a little overpriced and halflings a lot overpriced.

  30. This comment has been removed by the author.

  31. I have a copy of Dragon 109, and in early 2010 I blogged about my thoughts on the system after creating a couple of character classes. I found the numbers to be a a little off, and not always reflective of how attractive the class seemed on paper.

    I was going to mention OD&Dities #7, but the previous commenter beat me to it. I think that system, which is based on Dragon 109, is much better in terms of the math feeling right.

    If you want to see what I came up with from Dragon 109:
    My stab at custom classes

  32. I have a page that generates a random class based on this article. Even if you don't use it for PCs it might be useful for important NPCs:

  33. In the two campaigns where we used the Dragon article, it was an all-or-nothing approach, where we used the article only and ditched the canonical progression.

    I never played in a campaign that mixed them, so comparison never came up, really.

    I remember finding the article fairly mindblowing when I first read it, since I was still in those days very much a D&D player, so the idea of _creating_ a character (rather than just "rolling one up") hadn't sunk in yet.

  34. The Crabaugh article -- as bad as it is -- at least does not have negative experience point factors in it.

    In my own quixotic attempt at adapting the Crabaugh method, I tried to shy away from negative modifiers too (only using it once), it seems like a rather obvious design flaw in the 2E method.

  35. Anonymous12:10 PM

    I feel like the main classes are balanced against each other pretty perfectly: fighter, wizard and cleric. Dwarf has some advantages over fighters but that's balanced out by them being dwarves! What are they going to do, for example, if someone grabs them by the scruff of the neck? They're superior on paper only.

    Thieves at first glance, and at low levels are weak as crap, but check out their skills around level 10 or's 99% across the board...infallible sneaking, hiding, unlocking...yes please! (That doesn't happen in AD&D btw, it's a much slower progression.) Just think what you could do with that. He's Just a slow starting class like the MU.

    Elves definitely need the much higher XP requirement; they just do too much. Halflings...I've never really paid that much attention to so I can't rightly say how they balance out! I suspect they're on the weak side.

    The actual feels like there's something going on. MU and dwarf exactly identical XP amounts? Fighter + MU is 3600 even, exactly 80% of their combined btb value. I too would prefer to see rounder numbers in the 10s digit.

    But the whole exercise of class building feels completely wrongheaded. The random generator anarchist posted is creating bland mixes of existing classes. Why not just let your players play something actually fun and eyeball the balance? It's not like it's going out for publication or something, you'll have plenty of chances to tinker with it. Throw away the formula and have a little fun, next thing you know you'll be running the next Arduin.

  36. Great post!
    @ Erin
    I don't think Steven was "gaming the system." Zero cost should not be the baseline for abilities that increase with level, even if those abilities are comparatively the worst of all classes.

  37. @Chaz: I've gone back and tried some revisions that discard non-B/X abilities and eliminate zero/negative values.

    The canonical XP figures are elusive. You get Cleric to work, but the others are off by +/-100XP. Adjust them, and then Cleric is off. Rinse and repeat...

    Which suggests what we probably already know: the original XP values were based on "feel" instead of some consistent ability-to-XP correspondence. Put another way, devising a class-creation system that (1) works for the canon and (2) is devoid of flaws already pointed out may be an exercise in futility.

    Not that I've given up, mind you...

  38. I like this system, but I can also see how it could lead to point-buy.

  39. @ Walker, who said "But then I am not aware that any of the early designers really grasped a deep understanding of the mathematics behind the rules."

    That's ridiculous. Look at some of the design and analysis articles of the time. Most of the designers were coming hard-core off of tabletop wargame and grid-and-counter wargame design.

    Whether anyone at TSR knew what they were doing is another question, of course, but in general, the notion that the mathy guts of game design is some kind of new discovery is just flat out worthy of ridicule.

  40. I agree with your analysis for the most part. I always felt that Halflings got the short end of the stick (no pun intended)

    I happen to disagree with the Magic-User though. To me the psuedo-balance between them was that the Fighter progressed Linearly while the Magic-User progressed exponentially in power, this was reflected in their XP requirements.

    Now at first level it may seem overkill to for the Magic User to require 25% more than the Fighter to level but think for a minute how XP is commonly earned, Fighting monsters and collecting treasure.

    In both these situations the Fighter is doing most of the work, hacking bad guys and hauling around large sacks of coins. In effect he's doing most of the work, yet the system for the most part rewards each player equally.

    In that regard you can think of the XP to 2nd level as a measure of who is trying the hardest and putting forth the most exertion to get there. Certainly, not Phinneas V. Fire-and-forget Wizardton who is more than happy to coast while the meatheads do all the work.

    Out of curiosity, did Crabaugh provide XP requirements for anything past 2nd level?

  41. @RedHobbit, you're crazy! Phineas' T Wizardton's Sleep spell is the "get out of TPK free" card of B/X D&D.

    And don't even get me started about Charm Person as a force multiplier. The average enemy gets to save vs spells once per month! If you don't have a gang hench-monsters rolling with the party, you're doing it wrong.


  42. A spellcasting class like the cleric that can't cast spells at 1st level is bloody absurd.

    1. I always thought that made perfect sense. The Acolyte having to prove his faith and become an Adept before his Deity answers his prayers and grants him access to miracles.

  43. That clerics didn’t get spells at first level maybe says something about the original conception of the class. That they changed that in AD&D perhaps also says something about a change in that conception.

    In B/X, spells feel slightly less “core” to the class.